[v2,2/2] ACPI: HMAT: Fix initiator registration for single-initiator systems

Message ID 20221116-acpi_hmat_fix-v2-2-3712569be691@intel.com
State New
Headers
Series ACPI: HMAT: fix single-initiator target registrations |

Commit Message

Verma, Vishal L Nov. 16, 2022, 11:37 p.m. UTC
  In a system with a single initiator node, and one or more memory-only
'target' nodes, the memory-only node(s) would fail to register their
initiator node correctly. i.e. in sysfs:

  # ls /sys/devices/system/node/node0/access0/targets/
  node0

Where as the correct behavior should be:

  # ls /sys/devices/system/node/node0/access0/targets/
  node0 node1

This happened because hmat_register_target_initiators() uses list_sort()
to sort the initiator list, but the sort comparision function
(initiator_cmp()) is overloaded to also set the node mask's bits.

In a system with a single initiator, the list is singular, and list_sort
elides the comparision helper call. Thus the node mask never gets set,
and the subsequent search for the best initiator comes up empty.

Add a new helper to consume the sorted initiator list, and generate the
nodemask, decoupling it from the overloaded initiator_cmp() comparision
callback. This prevents the singular list corner case naturally, and
makes the code easier to follow as well.

Cc: <stable@vger.kernel.org>
Cc: Rafael J. Wysocki <rafael@kernel.org>
Cc: Liu Shixin <liushixin2@huawei.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reported-by: Chris Piper <chris.d.piper@intel.com>
Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
---
 drivers/acpi/numa/hmat.c | 26 ++++++++++++++++++++------
 1 file changed, 20 insertions(+), 6 deletions(-)
  

Comments

Kirill A. Shutemov Nov. 16, 2022, 11:54 p.m. UTC | #1
On Wed, Nov 16, 2022 at 04:37:37PM -0700, Vishal Verma wrote:
> In a system with a single initiator node, and one or more memory-only
> 'target' nodes, the memory-only node(s) would fail to register their
> initiator node correctly. i.e. in sysfs:
> 
>   # ls /sys/devices/system/node/node0/access0/targets/
>   node0
> 
> Where as the correct behavior should be:
> 
>   # ls /sys/devices/system/node/node0/access0/targets/
>   node0 node1
> 
> This happened because hmat_register_target_initiators() uses list_sort()
> to sort the initiator list, but the sort comparision function
> (initiator_cmp()) is overloaded to also set the node mask's bits.
> 
> In a system with a single initiator, the list is singular, and list_sort
> elides the comparision helper call. Thus the node mask never gets set,
> and the subsequent search for the best initiator comes up empty.
> 
> Add a new helper to consume the sorted initiator list, and generate the
> nodemask, decoupling it from the overloaded initiator_cmp() comparision
> callback. This prevents the singular list corner case naturally, and
> makes the code easier to follow as well.
> 
> Cc: <stable@vger.kernel.org>
> Cc: Rafael J. Wysocki <rafael@kernel.org>
> Cc: Liu Shixin <liushixin2@huawei.com>
> Cc: Dan Williams <dan.j.williams@intel.com>
> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> Reported-by: Chris Piper <chris.d.piper@intel.com>
> Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>

Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
  
Wysocki, Rafael J Nov. 17, 2022, 4:20 p.m. UTC | #2
On 11/17/2022 12:37 AM, Vishal Verma wrote:
> In a system with a single initiator node, and one or more memory-only
> 'target' nodes, the memory-only node(s) would fail to register their
> initiator node correctly. i.e. in sysfs:
>
>    # ls /sys/devices/system/node/node0/access0/targets/
>    node0
>
> Where as the correct behavior should be:
>
>    # ls /sys/devices/system/node/node0/access0/targets/
>    node0 node1
>
> This happened because hmat_register_target_initiators() uses list_sort()
> to sort the initiator list, but the sort comparision function
> (initiator_cmp()) is overloaded to also set the node mask's bits.
>
> In a system with a single initiator, the list is singular, and list_sort
> elides the comparision helper call. Thus the node mask never gets set,
> and the subsequent search for the best initiator comes up empty.
>
> Add a new helper to consume the sorted initiator list, and generate the
> nodemask, decoupling it from the overloaded initiator_cmp() comparision
> callback. This prevents the singular list corner case naturally, and
> makes the code easier to follow as well.
>
> Cc: <stable@vger.kernel.org>
> Cc: Rafael J. Wysocki <rafael@kernel.org>
> Cc: Liu Shixin <liushixin2@huawei.com>
> Cc: Dan Williams <dan.j.williams@intel.com>
> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> Reported-by: Chris Piper <chris.d.piper@intel.com>
> Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>

Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>


> ---
>   drivers/acpi/numa/hmat.c | 26 ++++++++++++++++++++------
>   1 file changed, 20 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/acpi/numa/hmat.c b/drivers/acpi/numa/hmat.c
> index 144a84f429ed..6cceca64a6bc 100644
> --- a/drivers/acpi/numa/hmat.c
> +++ b/drivers/acpi/numa/hmat.c
> @@ -562,17 +562,26 @@ static int initiator_cmp(void *priv, const struct list_head *a,
>   {
>   	struct memory_initiator *ia;
>   	struct memory_initiator *ib;
> -	unsigned long *p_nodes = priv;
>   
>   	ia = list_entry(a, struct memory_initiator, node);
>   	ib = list_entry(b, struct memory_initiator, node);
>   
> -	set_bit(ia->processor_pxm, p_nodes);
> -	set_bit(ib->processor_pxm, p_nodes);
> -
>   	return ia->processor_pxm - ib->processor_pxm;
>   }
>   
> +static int initiators_to_nodemask(unsigned long *p_nodes)
> +{
> +	struct memory_initiator *initiator;
> +
> +	if (list_empty(&initiators))
> +		return -ENXIO;
> +
> +	list_for_each_entry(initiator, &initiators, node)
> +		set_bit(initiator->processor_pxm, p_nodes);
> +
> +	return 0;
> +}
> +
>   static void hmat_register_target_initiators(struct memory_target *target)
>   {
>   	static DECLARE_BITMAP(p_nodes, MAX_NUMNODES);
> @@ -609,7 +618,10 @@ static void hmat_register_target_initiators(struct memory_target *target)
>   	 * initiators.
>   	 */
>   	bitmap_zero(p_nodes, MAX_NUMNODES);
> -	list_sort(p_nodes, &initiators, initiator_cmp);
> +	list_sort(NULL, &initiators, initiator_cmp);
> +	if (initiators_to_nodemask(p_nodes) < 0)
> +		return;
> +
>   	if (!access0done) {
>   		for (i = WRITE_LATENCY; i <= READ_BANDWIDTH; i++) {
>   			loc = localities_types[i];
> @@ -643,7 +655,9 @@ static void hmat_register_target_initiators(struct memory_target *target)
>   
>   	/* Access 1 ignores Generic Initiators */
>   	bitmap_zero(p_nodes, MAX_NUMNODES);
> -	list_sort(p_nodes, &initiators, initiator_cmp);
> +	if (initiators_to_nodemask(p_nodes) < 0)
> +		return;
> +
>   	for (i = WRITE_LATENCY; i <= READ_BANDWIDTH; i++) {
>   		loc = localities_types[i];
>   		if (!loc)
>
  

Patch

diff --git a/drivers/acpi/numa/hmat.c b/drivers/acpi/numa/hmat.c
index 144a84f429ed..6cceca64a6bc 100644
--- a/drivers/acpi/numa/hmat.c
+++ b/drivers/acpi/numa/hmat.c
@@ -562,17 +562,26 @@  static int initiator_cmp(void *priv, const struct list_head *a,
 {
 	struct memory_initiator *ia;
 	struct memory_initiator *ib;
-	unsigned long *p_nodes = priv;
 
 	ia = list_entry(a, struct memory_initiator, node);
 	ib = list_entry(b, struct memory_initiator, node);
 
-	set_bit(ia->processor_pxm, p_nodes);
-	set_bit(ib->processor_pxm, p_nodes);
-
 	return ia->processor_pxm - ib->processor_pxm;
 }
 
+static int initiators_to_nodemask(unsigned long *p_nodes)
+{
+	struct memory_initiator *initiator;
+
+	if (list_empty(&initiators))
+		return -ENXIO;
+
+	list_for_each_entry(initiator, &initiators, node)
+		set_bit(initiator->processor_pxm, p_nodes);
+
+	return 0;
+}
+
 static void hmat_register_target_initiators(struct memory_target *target)
 {
 	static DECLARE_BITMAP(p_nodes, MAX_NUMNODES);
@@ -609,7 +618,10 @@  static void hmat_register_target_initiators(struct memory_target *target)
 	 * initiators.
 	 */
 	bitmap_zero(p_nodes, MAX_NUMNODES);
-	list_sort(p_nodes, &initiators, initiator_cmp);
+	list_sort(NULL, &initiators, initiator_cmp);
+	if (initiators_to_nodemask(p_nodes) < 0)
+		return;
+
 	if (!access0done) {
 		for (i = WRITE_LATENCY; i <= READ_BANDWIDTH; i++) {
 			loc = localities_types[i];
@@ -643,7 +655,9 @@  static void hmat_register_target_initiators(struct memory_target *target)
 
 	/* Access 1 ignores Generic Initiators */
 	bitmap_zero(p_nodes, MAX_NUMNODES);
-	list_sort(p_nodes, &initiators, initiator_cmp);
+	if (initiators_to_nodemask(p_nodes) < 0)
+		return;
+
 	for (i = WRITE_LATENCY; i <= READ_BANDWIDTH; i++) {
 		loc = localities_types[i];
 		if (!loc)